pvyas96/Thin_Section_Dataset

Description

This dataset contains thin section images of 19 rocks and minerals compiled from various open-source websites. It is intended for research and educational purposes, particularly in the field of petrology. The dataset is sufficient to train and understand petrological problems in data science. Users can later expand the classes and data using any framework of their choice.

Sources

The data were collected from various open-source websites such as Mendeley Data, Digital Rocks Portal, datadryad.org, and Science Direct.

How to Use

Please refer to the provided notebooks at https://www.kaggle.com/code/prateekvyas/petronet-with-fastai-and-pytorch for examples on how to use this dataset for training deep learning models in petrology.

For any questions or assistance, feel free to comment on the dataset page or contact the author directly.

Application

App developed using this dataset is available at https://huggingface.co/spaces/pvyas96/thin_section_prediction